The Agents vs Zombies event is going great but our devs are running out of credits. No worries heres my guide to running up your own Local AI and Integrating with Outsystems with credit limits!!!

To do this we need to achieve 3 things

• Run a local AI model
• Expose an LLM end point
• Connect Outsystems to the same

Run a local AI model

To do this we need a Model Runner Ollama is the lightest and easiest to setup.

Download and install the setup of your choice from the below URL

https://ollama.com/download/windows

Once setup open Ollama and select a model of your choice. I would start with gemma3:4b as it will fit in most laptops with 8 to 16 GB Ram.

Once Selected enter a sample query “ just respond Yes”

If this is the first time you are using Ollama or the model it will download and load the model for you.

Once you get a response it means your model is now ready for use

Now open the side bar and click on settings

Ensure the Expose Olamma to the network is enabled

Now your Ollama is set up and ready to be exposed as an end point.

Exposing an Endpoint

To do this we can use Ngrok that give you one free end point for development.

Go to the below URL signup and download your installer of choice.

https://ngrok.com/download/windows

Once Set up is done  go to the below url and get your authorisation token

https://dashboard.ngrok.com/get-started/your-authtoken

Go to https://dashboard.ngrok.com/domains

Click on + new domain and get a Free domain generated, copy and save your domain in a notepad

Now open power shell on windows

And enter the below command replace $YOUR_AUTHTOKEN with your actual token this will register the instance

ngrok config add-authtoken $YOUR_AUTHTOKEN

Now to initiate it you need to use the below command

ngrok http 11434 --host-header="localhost:11434"

On running the command if it all went well you should see the below screen

Now your LLM end point is ready!!

Connecting A custom LLM

Go to your personal ODC environment and click on AI models

Click on Add An AI model

Select a custom AI model

Add your Connection name and description

Then click on add endpoint

Enter your end point details as below

Name:

Any name you want

Model ID:

Use the Model id in Ollama here I am using “gemma3:4b”

URL:

Replace "your-domain-name" with your domain name

“https://your-domain-name/v1”

And click on test end point you should get a success message now click on save.

And now save the connection

Congratulations you have now connected your local LLM to Outsystems

After this the rest of the process is well explained in the link below by Bruno Martinh

https://www.linkedin.com/feed/update/urn:li:activity:7389263711133315072/

Happy Hunting!!!!